regression: radeon DRI and SMP is bad

Bug #16686 reported by Johan Walles
10
Affects Status Importance Assigned to Milestone
xserver-xorg-video-ati (Ubuntu)
Fix Released
Medium
Unassigned

Bug Description

I've just upgraded my four CPU i386 system to Hoary.

If I boot with an SMP kernel, X freezes about 30 seconds after I have logged in
(thus Severity: critical == crashes, data loss). I haven't (yet) seen any
freeze at the GDM screen.

If I boot with a single-CPU kernel, X works fine.

When X is frozen, neither ctrl-alt-f1 nor ctrl-alt-backspace gets me out; I have
shut down my system each time this happens.

lspci says that I have this graphics card:
0000:01:00.0 VGA compatible controller: ATI Technologies Inc Radeon RV100 QY
[Radeon 7000/VE]

Will attach my /etc/X11/xorg.conf.

Revision history for this message
Johan Walles (walles) wrote :

Created an attachment (id=2207)
My xorg.conf

I haven't touched my xorg.conf, I assume it was generated when I upgraded to
Hoary.

Revision history for this message
Johan Walles (walles) wrote :

johan@transwarp:~$ dpkg -l xserver-xorg
[...]
||/ Namn Version Beskrivning
+++-==============-==============-============================================
ii xserver-xorg 6.8.2-10 the X.Org X server

Revision history for this message
Daniel Stone (daniels) wrote :

Any chance you could give this a go with breezy?

Revision history for this message
Johan Walles (walles) wrote :

If there's a live CD for Breezy with SMP kernels on it, yes, otherwise no. Are
there any of those? The Hoary live CD had only non-SMP kernels.

This is my workstation at work, so I really don't want to mess around with it
too much. A live CD would be peachy though.

Revision history for this message
Daniel Stone (daniels) wrote :

Not yet, unfortunately, but thanks for following up. :) If you could give this a
shot when it's more convenient for you, that would be fantastic. Thanks.

Revision history for this message
Johan Walles (walles) wrote :

I'll try this out either when an SMP Live CD comes out (bug 12622), or shortly
after Breezy is released.

Revision history for this message
Matt Zimmerman (mdz) wrote :

If you can't test this before the release, we won't be able to fix it. Can you
install Breezy on a second partition for testing purposes?

Revision history for this message
Johan Walles (walles) wrote :

I'm sorry, no.

I realize this means you won't be able to do anything about it, at least before
release. That's unforturnate, but this isn't anything I can put too much time into.

I'll let you know what happened after I upgrade to Breezy, or when the SMP live
CD comes out, whichever comes first.

Revision history for this message
Matt Zimmerman (mdz) wrote :

(In reply to comment #8)
> I'm sorry, no.
>
> I realize this means you won't be able to do anything about it, at least before
> release. That's unforturnate, but this isn't anything I can put too much time
into.
>
> I'll let you know what happened after I upgrade to Breezy, or when the SMP live
> CD comes out, whichever comes first.

Unfortunately, if we need to wait 6 months for you to test a new package, it is
unlikely that this bug will be fixed in less than one year. Lowering severity
since it can't be confirmed in recent code.

Revision history for this message
Cedric Veilleux (cveilleux) wrote :

I have the same graphic card:

0000:01:00.0 VGA compatible controller: ATI Technologies Inc Radeon RV100 QY
[Radeon 7000/VE]

On an Hyper-Threading P4. If I boot an SMP kernel, the system freezes about 30
seconds after I login an X Session, but it is perfectly stable with a non-smp
kernel.

I have this problem in Hoary *and* in the latest Breezy (tried yesterday).

I am using the 'ati' X driver. I will try the fglrx one and give some feedback
later.

Revision history for this message
Cedric Veilleux (cveilleux) wrote :

When loading the fglrx kernel module, I get the following error:

No device found!

So I have been unable to test X with the fglrx driver instead of the ati one. I
did took the time to put the SMP kernel under some stress using cpuburn and it
seemed very stable, as long as X was not running.

Revision history for this message
Hector (h-centeno) wrote :

(In reply to comment #9)
> (In reply to comment #8)
> > I'm sorry, no.
> >
> > I realize this means you won't be able to do anything about it, at least before
> > release. That's unforturnate, but this isn't anything I can put too much time
> into.
> >
> > I'll let you know what happened after I upgrade to Breezy, or when the SMP live
> > CD comes out, whichever comes first.
>
> Unfortunately, if we need to wait 6 months for you to test a new package, it is
> unlikely that this bug will be fixed in less than one year. Lowering severity
> since it can't be confirmed in recent code.

I have the same problem and I'm using Breezy. Whenever I use the smp kernel in
combination of my Radeon 7000 card the system gets completely frozen. If I
disable "dri" in my xorg.conf then everything is ok (except that I don't have
any 3d acceleration, of course). This is a common bug and I found a reference to
it here: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=152648

I hope this can be fixed for Breezy.

Revision history for this message
Hector (h-centeno) wrote :

... and here there is more information about it that might be helpful, a little
bit outdated though.
http://lists.freedesktop.org/pipermail/xorg/2004-November/004733.html

Revision history for this message
Matt Zimmerman (mdz) wrote :

(In reply to comment #12)
> I have the same problem and I'm using Breezy. Whenever I use the smp kernel in
> combination of my Radeon 7000 card the system gets completely frozen. If I
> disable "dri" in my xorg.conf then everything is ok (except that I don't have
> any 3d acceleration, of course). This is a common bug and I found a reference to
> it here: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=152648

Thanks for testing.

> I hope this can be fixed for Breezy.

That's unlikely unless the fix is extremely simple and obvious. Ubuntu 5.10
releases in just a few days, and there is no time to regression test non-obvious
fixes.

Revision history for this message
Hector (h-centeno) wrote :

(In reply to comment #14)
> (In reply to comment #12)
> > I have the same problem and I'm using Breezy. Whenever I use the smp kernel in
> > combination of my Radeon 7000 card the system gets completely frozen. If I
> > disable "dri" in my xorg.conf then everything is ok (except that I don't have
> > any 3d acceleration, of course). This is a common bug and I found a reference to
> > it here: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=152648
>
> Thanks for testing.
>
> > I hope this can be fixed for Breezy.
>
> That's unlikely unless the fix is extremely simple and obvious. Ubuntu 5.10
> releases in just a few days, and there is no time to regression test non-obvious
> fixes.

...well, whatever time it takes I'm willing to help. Just let me know what can I
do to contribute make things clear for you (e.g. testing, sending logs, etc). On
my side I'll keep trying to find what is this about.

thanks!

Revision history for this message
Hector (h-centeno) wrote :

(In reply to comment #15)
> (In reply to comment #14)
> > (In reply to comment #12)
> > > I have the same problem and I'm using Breezy. Whenever I use the smp kernel in
> > > combination of my Radeon 7000 card the system gets completely frozen. If I
> > > disable "dri" in my xorg.conf then everything is ok (except that I don't have
> > > any 3d acceleration, of course). This is a common bug and I found a
reference to
> > > it here: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=152648
> >
> > Thanks for testing.
> >
> > > I hope this can be fixed for Breezy.
> >
> > That's unlikely unless the fix is extremely simple and obvious. Ubuntu 5.10
> > releases in just a few days, and there is no time to regression test non-obvious
> > fixes.
>
>
> ...well, whatever time it takes I'm willing to help. Just let me know what can I
> do to contribute make things clear for you (e.g. testing, sending logs, etc). On
> my side I'll keep trying to find what is this about.
>
> thanks!

I found a discussion about this bug including some patches:
https://bugs.freedesktop.org/show_bug.cgi?id=1912

hector

Revision history for this message
Daniel Stone (daniels) wrote :

we'll pick this one up when we get the new ati driver in dapper

Revision history for this message
Johan Walles (walles) wrote :

Daniel Stone wrote on 2005-10-18: "we'll pick this one up when we get the new ati driver in dapper".

I'm not sure what that means exactly, but since Dapper is out and that was almost a year ago, I'm changing status from Rejected -> Unconfirmed, hoping for re-triage.

This is a regression from Warty.

Changed in xorg:
status: Rejected → Unconfirmed
Revision history for this message
ea (alyaz) wrote :

I'm using Dapperdrake with
kernel 2.6.15-23-686 with SMP.
X Window System Version 7.0.0
ATI Radeon RV100 QY [Radeon 7000/VE]

System freezes when dri is enabled in xorg.conf (as described in this bug)
It seems like this bug is rv100-specific.

Please reopen.

Revision history for this message
ea (alyaz) wrote :
Daniel Stone (daniels)
Changed in xorg:
assignee: daniels → ubuntu-x-swat
Revision history for this message
Steven Harms (sharms) wrote :

Is this still an issue on Feisty?

Changed in xorg:
status: Unconfirmed → Needs Info
Revision history for this message
Johan Walles (walles) wrote :

During the almost two years since I first reported this problem I've replaced my hardware, so that ATI card is no more with me. Sorry, cannot answer the question.

Some of the other people on this bug might though?

Revision history for this message
w4ett (w4ett) wrote :

Steven : seems it is still an issue in Feisty See: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-ati/+bug/114520

Changed in xorg:
status: Needs Info → Confirmed
Revision history for this message
Timo Aaltonen (tjaalton) wrote :

Nope, that bug is for another model, this one was about rv100 and it should be fixed long ago.

Changed in xorg:
assignee: ubuntu-x-swat → nobody
status: Confirmed → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Bug attachments

Remote bug watches

Bug watches keep track of this bug in other bug trackers.